The theory is if one uses Firefox a lot, re-boot periodically (e.g. every 24 hrs) because of memory leakage.

Maybe there is already some good evidence pro or con.

I am not a technical person. But, I started re-booting all five of my computers every morning because I felt I needed to do two of them any way, three year old Vista machines. Both were slowing down on loading second and third web pages. Re-booting solved that.

The two Vista machines are maxed at 4 gigs of DDR2 single channel; two Win 7's have 4 gigs and 6 gigs respectively of DDR3 double channel. The last machine, a real trooper, is my netbook an Atom N270, 1 gig of DRAM, just on Spinhenge and Leiden and HCC at WCG, stuff it can handle, never yet a failure, it even completed a valid "A" WU at DDDT2.

So, I noticed that my productivity went up about 1000 points a day after I started re-booting. I know that I wait for others to finish and/or they wait for me. So, now I am averaging around 5000 per day.

So, I started yesterday, not re-booting, and, bingo, today I am getting to just about 4000 late in the 24 hour period.

Unless someone tells me otherwise, I am going to let this run for about three more days and see what happens, and then go back to re-booting every morning.

I don't care about giving up credit if there is something to be learned.

But, if I am way off base, please let me know.

---

http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings

The theory is if one uses Firefox a lot, re-boot periodically (e.g. every 24 hrs) because of memory leakage.

Maybe there is already some good evidence pro or con.

https://bugzilla.mozilla.org/show_bug.cgi?id=545378

:-D

---

So, hey, Jord, how are you?

So you confirm the leakage business, and even I assume that is you at Comment 3.

The question is, does this memory leakage affect crunching? I check my stats at boincstats.com (user is "mitrichr"). I never cleared 4000 yesterday, my average is over 5000. So, I am going today and tomorrow with no re-boot. Then I will re-boot.

I mean, you know, I love FF. So if it is a problem, solved by this work-around, then that's fine.

---

http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings

The question is, does this memory leakage affect crunching?

Only if you leave FF open on a page that has a news ticker, or something like that. My maximum memory usage at one point (not registered there) was over 500MB of RAM. Then some tasks may stop and show in BOINC Manager as Waiting for memory.

---

Thanks, Jord, I think that I will play this thing out until Thursday morning and then re-boot all machines.

If my results by Saturday then are at all instructive, I will come back here and comment.

---

http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings

By the way, it's not necessary to reboot. A Windows log off and log on will do just as nicely to clear up memory. :-)

---

Thanks.

---

http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings

Theory going up the stack? 7000 credits today.

I did need to re-boot one of the Vista machines, having nothing to do with BOINC.

---

http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings

Theory going up the stack? 7000 credits today.

Do you add the pending credits into that sum too?

Gruß,
Gundolf

---

Computer sind nicht alles im Leben. (Kleiner Scherz)

Theory going up the stack? 7000 credits today.

Do you add the pending credits into that sum too?

Gruß,
Gundolf

No, I get my results from BOINC Stats. if you want a look, the user id is mitrichr.

---

http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings

mitrichr,

Start fetching BOINCTasks which shows per-job CPU efficiency and make it a routine to minimize Firefox. That leaking and growing use of RAM and VM is not going away. Been with FFX from the beginning though now it's much less. Minimizing actually quite quickly knocks down the memory use at the same time.

Booting is like restarting the client. You'll loose at least the time between last checkpoint and point of shut down. On an 4 core HT and 8 FAAH jobs running, that's probably in excess of 1 hour CPU time gone, just like that.

Then there is that new sneaky feature of Whilst my computer gently wheeps... suspending computation each time overall non-BOINC CPU use exceeds 25%. Even at 100% it kicks in when running based on prefs and does a 4 second plus pause, so set that with a > 6.10.33 client to Zero.

Just a few notes.

Thanks, Sek -

Please explain "fetching BOINCTasks". I am on 6.10.43, and "Whilst' is set at zero on all.

---

http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings

Theory going up the stack? 7000 credits today.

Do you add the pending credits into that sum too?

No, I get my results from BOINC Stats. if you want a look, the user id is mitrichr.

But then your theory stands on very weak feet anyway, since the amount of pending credit can change vastly from day to day. That was probably the reason for the sudden 7000 value too.

Gruß,
Gundolf

But then your theory stands on very weak feet anyway, since the amount of pending credit can change vastly from day to day. That was probably the reason for the sudden 7000 value too.

Gruß,
Gundolf

I do know about the day-to-day variability in results, as different hosts will take different amounts of time to finish. My "theory" was posed as a question, so I will not be surprised at any results. I just think that I saw a pattern emerge after I began re-boots every 24 hours.

I am sure Sek is correct about the loss of time spent.

I am still going to play it out one or two more days, maybe until Friday morning. Thanks for your considerations.

---

http://sciencesprings.wordpress.com
http://facebook.com/sciencesprings